Our long-term programmatic goal is to develop screening strategies to diagnose asymptomatic pancreatic cancer (PaC). Up to 80% of PaCs have hyperglycemia and diabetes (DM), which is evident many months prior to the cancer diagnosis and improves following resection of PaC. Conversely, older subjects with newonset DM have a ~8 fold higher risk of having PaC compared to the general population. Recognition of new-onset DM as an early manifestation of PaC could lead to diagnosis of asymptomatic early stage PaC. In this proposal we take our strong and consistent clinical and epidemiological observations to the laboratory to understand the pathogenesis of PaC-associated DM (PaCDM) and identify its biomarkers. Specific Aim 1: To determine if B-cell dysfunction is an early and key defect in PaCDM: DM occurs in insulin resistant states when B-cells fail to compensate for impaired insulin action. The very high prevalence of DM in PaC implies high rate of B-cell failure. We have developed a technique in humans to simultaneously assess B-cell function, insulin sensitivity, and hepatic insulin extraction using 3 radiolabeled glucose tracers. We have used this technique to study subjects with type 2 DM, impaired glucose tolerance and normal glucose tolerance. We will perform similar studies in PaC to determine if B-cell dysfunction is an early and key defect in glucose metabolism in PaCDM. Specific Aim 2: To determine if adrenomedullin (AM) is the mediator of PaCDM: AM is a 52 amino acid peptide hormone expressed in normal human islets that inhibits insulin exocytosis from B-cells. It is markedly overexpressed in PaC and its plasma levels are increased in PaCDM. We hypothesize thai AM is the mediator of B-cell dysfunction in PaC. In preliminary studies we have been able to "transmit" DM from a human to SCID mice using a xenograft of PaC from a patient with PaCDM, while a xenograft of PaC from a patient with normal fasting glucose had no effect on glucose levels. In in vitro studies using INS-1, an insulinoma cell line, we have observed that PaC cell lines inhibit glucosemediated insulin release. Using genetic and pharmacological methods to modulate the expression and action of AM in these in vivo and in vitro models, we will investigate the role of AM in causing PaCDM. Specific Aim 3: To develop a predictive model for PaC among new-onset diabetics: PaCDM is associated not only with high plasma AM levels, but also with older age, obesity and family history of DM. We will measure plasma AM, insulin and glucose levels as well as CA 19-9, the best known tumor marker of PaC, in a large cohort (n=420) of subjects with PaC with and without new-onset DM, new-onset type 2 DM, and healthy and disease controls. We will determine the performance characteristics of AM as a biomarker of PaCDM and develop a predictive model for PaC using laboratory, clinical and demographic predictive factors. Translational Significance: If a biomarker of PaC-induced DM is identified it will have immediate clinical impact as it will allow us to start screening for asymptomatic PaC in the subjects with new-onset DM.